Comparative Analysis on Skin Cancer diagnosis using Deep Learning Neural Networks

Abstract

The skin can be considered as the largest part in the human being .Majorly skin cancer can be classified into two types like melanocytic nature and non melanocytic nature. Among these two melanocytic is more dangerous due to spreading nature under epidermis and sub-cutaneous layers. The dermoscopic methods are time consuming process in early prediction of melanocytic lesions. There are many deep learning neural network algorithms were employed for early prediction of skin cancer. During the prediction of the skin condition different factors are considered like artifacts, hair, color and noise .In skin cancer diagnosis, pre-processing steps and augmentation techniques were employed. The mostly used algorithm is deep convolutional neural network (DCNN).This paper proposes intelligent hybrid deep learning algorithms (IHDLA) for improving the prediction accuracy.